
TM1300 Data Book
Philips Semiconductors
14-16
PRODUCT SPECIFICATION
up scaled by 2 relative to the Y component for YUV 4:4:4
output, although they could be up scaled as part of gen-
eral up scaling of the image.
The YUV 4:2:2 output mode also provides higher pro-
cessing bandwidth relative to YUV 4:4:4 up scaling. Half
as many U and V pixels are processed.The output pixel
rate is one pixel per 20 nanoseconds for the YUV 4:2:2
output mode versus one pixel per 30 for conversion to
YUV 4:4:4. This can be used to provide some processing
performance improvement for very large images at the
expense of some chroma quality.
14.5.9.2
PCI output block timing
The ICP outputs pixels to the PCI interface at a peak rate
of 33 Mpix/sec in RGB mode and 50 Mpix/second in the
YUV mode using YUV sequencing. For one word per pix-
el output codes, such as RGB-24, this is a peak rate of
33 Mwords/sec or 132 Mpix/sec in the RGB sequencing
mode. This is the same speed as the 132 MB/sec peak
rate of the PCI interface. (At 50 Mpix/sec, the result
would be 200 MB/sec.) The BIU control for the PCI inter-
face has a FIFO for buffering data from the ICP, but this
buffer is only 16 words deep. Therefore, the ICP will oc-
casionally have to wait for the PCI to accept more data.
In the PCI output mode, this stalls the ICP clock.
14.6
OPERATION AND PROGRAMMING
The ICP uses a combination of hardware and a Micro-
program Control Unit (MCU) to implement its scaling, fil-
tering and conversion functions. The microprogram is a
T
o
PCI
5 Stage Multiplier-
Accumulator
Y, U, V LSBs
Reg
a
+2
RAM
a
+1
RAM
a
+0
RAM
a
-1
RAM
a
-2
RAM
Y Counter
Y LSB Counter
Buffers 0,1
Block FIFO
Filter Source Select
5-tap Filter
Reg
U Counter
U LSB Counter
Buffers 2,3
Block FIFO
Reg
V Counter
V LSB Counter
Buffers 4,5
Block FIFO
Reg
OL Counter
B, BX Counter
Buffer 8
Bit Mask
Buffers 6,7
Overlay
FIFO
Multiple
x
er:
Y
,U
,V
Select
Mux
YUV
to
RGB
Con
ver
sion,
Formatting,
Alpha
Blending
&
Bit
Masking
YUV
Counter
Sequence
Pixel
Clock
Y, U, V Data FIFO Clocks
Mirr
or
Multiple
x
e
r
Y Mirror Cntr
U Mirror Cntr
V Mirror Cntr
Mux
RGB to SDRAM case
RGB to PCI case
Figure 14-16. ICP horizontal scaling for RGB output data flow block diagram